The morphology-syntax interface

and simple X-bar schemata allows Chomsky to account for the syntactic parallelisms between these three types of expressions (verbs, DNs and GNs) in a uniform way. The idea that some DNs belong in the Lexicon rather than the syntax came to be known as the lexicalist hypothesis to derivational morphology. There are two theoretical positions within this lexicalist approach to derivational processes, which are conceptually different, though often not distinguished: (i) what is generally known as the weak lexicalist hypothesis, by which DNs are mostly lexically derived, . but which could admit some transformational derivations of D~s, and (li) what Perlmutter (1988) refers to as the 'split morphology' hypothesis, which denies the possibility that there are DNs that can be derived by means of transformations.s Some of the works dealing with these issues can be understood as advocating one or the other position. In this regard, Chomsky's "Remarks" can be read in either of the two ways. The same applies to Anderson's (1982) work, in which what is derivational and what is inflectional is not independently characterized: those processes which happen to be syntactically relevant are inflectional and those which are not, in contrast, are derivational and take place in a morphological component; but what is inflectional in one language could be taken to be derivational in another language and viceversa (see also Lieber 1992). The lexicalist hypothesis which emerged from "Remarks" paved the way, finally, for another interpretation of the syntax-morphology relation, by which morphological operations in general -whether they are inflectional or derivationaltake place in the Lexicon: i.e. transformational rules cannot refer to wordinternal processes. This is known as the strong lexicalist hypothesis, whose origins are found in J ackendoffs (1972) (Extended) Lexicalist HYpothesis) and which gained strong support in the 70s and 80s (see e.g. Lapointe 1980, 1988). Some advocates of the strong lexicalist hypothesis like Selkirk (1982) and Di Sciullo and Williams (1987), however, allow syntactic rules to refer to morphological features. The strong lexicalist hypothesis underlies to a wider or a lesser extent several proposals concerning the relation between morphology and phonology (as well as other components of the grammar), to which we now turn. The study of the interaction between morphological and phonological operations during the 70s gave rise to a more elaborated theory of word formation processes as well as to a more sophisticated view of the structure of the Lexicon in what is known as the Level Ordering HYpothesis, by which the Lexicon is divided into a series (5) For related discussion see Hendrick (1995) and references therein. XlV AMAYA MENDIKOETXEA & MYRIAM URIBE-ETXEBARRIA of levels or strata, each with its own set of affixes (Allen 1978, Siegel 1979).6 This idea was further developed in the early 80s in the framework that came to be known as the Theory of Lexical Phonology and Morphology, originally put forth by Kiparsky (1982), which presented a detailed theory of a level-ordered Lexicon on the basis of the interaction between phonological and morphological processes. Kiparsky, on the basis of work carried out by Mascaro (1976) and Pesetsky (1979) concerning the cyclic application of phonological rules, argued that cyclicity effects in phonology could be straightforwardly derived by appealing to a level ordered Lexicon, along the lines of the Level Ordering Hypothesis. Within this approach, each of the levels into which the Lexicon is divided contains a set of morphological rules and a related set of phonological operations. Whenever a morphological operation of a given level takes place, the output of this word formation operation is submitted to the set of phonological rules that are associated with that lexical level. Within this model, Kiparsky establishes a clear cut distinction between (i) phonological rules that apply (at one or more levels) in the Lexicon, and (ii) phonological rules that apply after words have been inserted into syntactic structures. The former, which he refers to as the rules of lexical phonology, are "intrinsically cyclic because they reapply after each step of word-formation at their morphological level"; the latter, which he refers to as the rules of postlexical phonology, are "intrinsically noncyclic" (op. cit.: 131-2).7 Unlike the rules of lexical phonology, the rules of postlexical phonology may apply word-internally as well as across word-boundaries; they are not affected by the internal structure of words or by the nature of the internal components of the word in which they apply. The idea of a level ordered Lexicon allowed Kiparsky to approach some questions in relation to the possible ordering of affixes (the order of affixes is determined by what level they belong to), as well as regarding existent and inexistent forms and blocking effects, providing a partial answer to some of the questions previously posed by Halle (1973). This approach to the Lexicon was further developed in the work of Halle and Mohanan (1985) and Mohanan (1982, 1986), and enjoyed a great deal of success during the eighties. The proposal defended in some of these works that (some aspects of) inflectional morphology -such as verbal inflection and Casewere located in the Lexicon provided support for a strong lexicalist hypothesis and for the view that morphology was a lexical phenomenon.s (6) Siegel (1979) distinguishes between two types of derivational affixes on the basis on Chomsky & Halle's (1968) distinction between two types of morphological boundaries and argues that the Lexicon should be divided intotwo blocks, each containing one class of affixes. Allen (1978) analyzes compounding and inflectional affixes, as well as derivational affixes, and argues that the Lexicon should be divided into fow: strata, each with its own set of rules: level I and level II for the two types of derivational affixes, level III for compounding and level IV for inflectional morphology. (7) While Kiparsky (1982) takes the rules of lexical phonology to be cyclic, this view is not shared by Mohanan & Mohanan (1984), who argued that all lexical strata in Malayalam are not cyclic, or by Halle & Mohanan (1985), who argued that Stratum 2 in English is not cyclic. See Hualde (1988), also within the framework of Lexical Phonology for arguments, that in Basque the phonological rules of a given level do not have a chance to apply every time a morphological operation of that Level applies but rather only once, after all the morphological processes associated with that level have taken place. For the proposal that lexical strata may be either cyclic or non-cyclic, see Halle & Mohanan (op. cit.) and Mohanan (1986). (8) See, among others, Sproat (1985) and Fabb (1988) for an overview of the issues that cast doubt on Lexical Phonology in the mid and late 80s. THE MORPHOLOGY-SThlTAX INTERFACE xv The incorporation of lexicalist hypotheses, in their different conceptions, into the grammar has led to new ways of approaching the relation between morphology, syntax and phonology. Regarding the model of the grammar, the consequences of adopting one or the other hypothesis are vast. If the split morphology hypothesis is adopted, the relation between morphology and syntax is restricted to processes dealing with inflectional morphology. If the strong lexicalist hypothesis is adopted, the relation between syntax and morphology is necessarily limited. The weak lexicalist hypothesis allows for a wide variety of morphological operations in the syntax, whether they are derivational or inflectional (with anything idiosyncratic and unproductive restricted to the Lexicon). Linguists working within the weak lexicalist hypothesis vary in the role they attribute to syntactic principles in accounting for morphological operations; they are divided into those who believe that specific morphological principles are still required (cf. Baker 1988) and those for whom syntactic principles suffice to account for morphological operations (c.f. Lieber 1992). In the next sections, we deal with some current issues concerning the relation between syntax and morphology in the late 80s and 90s, which are of direct relevance for the topics dealt with in the different papers in this volume. 1.2. Morphological operations and complex word formation in the Principles and Parameters model One of the basic questions concerning the relation between syntax and morphology is how to provide a structural representation for morphologically complex words. From the early 80s, there have been proposals in the literature in favor of generating the morphological structure of complex words by means of X-bar principles, along the lines proposed for syntactic structure (see e.g. Selkirk 1982 and Lieber 1992). Williams (1981), in particular, argued that words, like phrases, are headed, with the head as the rightmost morpheme of the complex form (the Righthand Head RuleJ.9 In addition to derivational affixes, inflectional affixes started to be analyzed as heads projecting their own phrases during the 80s. This view gained further support with the development of an approach where the status of heads was granted to functional elements such as Det, Tense, Comp and so on, some of which often show up as inflectional affixes in many languages. The idea that inflectional affixes are generated in· independent syntactic positions is already present in early work in generative grammar. However, it is not until the 80s that functional heads like C(omplementizer) and I(nflection) are assimilated into the X-bar theoretic framework as elements heading their own projections: CP and IP. These projections, introduced by Chomsky (1986) in Barners, replace the earlier S' and S categories, respectively, thus changing some of the previous assumptions regarding clausal architecture. At the same time, functional XPs ar